PR 23551: why should we coalesce inlined variables?

[Daniel Berlin]

On 7/9/07, Alexandre Oliva <aoliva@redhat.com> wrote:
> On Jul  8, 2007, "Daniel Berlin" <dberlin@dberlin.org> wrote:
>
> > Just so we are straight here, o the ones who debug at -O2 -g are
> > normal, but the people who make code like the testcase are  unusual.
> > :)
>
> If you still think debug information is something we emit to make the
> use of GDB more bearable, let me welcome you to the 21st century.

I have a simple answer to every single thing you have posed below
"Suggest doubling the memory usage for compile time and see what you
get back".

I work on exactly all of the problems below, and produce debug
information consumers that do all these.

I've also worked heavily in the past on producing good debug
information for optimized code (see var-tracking, dwarf2 location
lists, etc), so I find your comments both humorous and insulting at
the same time.

You are throwing pebbles into a lake here with this patch.

>
> For the testcases I was given before, the memory his was negligible.
> Is it just me or are you trying to overgeneralize the memory impact?

No.
You are trying to make it seem like "oh, well, we have a broken
testcase".  Doubling memory usage is never acceptable on almost any
testcase. Period.
You doubled memory usage on other Google internal testcases I have as well.
Please revert the patch, as you've been asked to by multiple people.

There are other ways to solve this problem.  Fix the underlying
problem of how we collect debug info, stop trying to hack
optimizations one by one.